| 摘要 |
PROBLEM TO BE SOLVED: To accurately read aloud the KANA(Japanese syllabary) reading of the same notation and different reading word preferentially with the reading used by a user in voice rule synthesis for reading an optional text aloud. SOLUTION: For the same notation and the different reading word, in the case that the user wants the desired KANA reading, by inputting the KANA reading to a KANA/KANJI(Chinese character) conversion part 101, the KANA/ KANJI conversion part 101 uses a KANA look-up index 103 and a thesaurus 105, detects the word No of the desired KANA reading for the word of the same notation and the different reading and stores it in a converted result storage part 106. When the same notation and the different reading word is present in the input text of a voice rule synthesis part 102, the voice rule synthesis part 102 extracts the word No of the word of the same notation and the different reading from a notation look-up index, and when the word No is present in the converted result storage part 106, detects the KANA reading by the word No from the term dictionary 105 and outputs it as voice. |
